Application On Face Recognition Using Class-Specific Hidden Markov Model Algorithm

Face recognition is one of the most challenging problems in the fields of patternrecognition and machine vision. Face recognition based on hidden Markov modelessentially uses the statistical method to create a mathematical model for a face image todistinguish each face image.The class-specific method was recently developed as a method of dimensionality reduction inclassification. Unlike other methods of dimension reduction, it is based on sufficient statistics andresults in no theoretical loss of performance. The class-specific method assign a separate feature setto each class. It’s feasible to extend the idea further to the problem of HMM modeling when eachstate of the HMM may have its own approximate sufficient statistic information.For the shortcomings that the effectiveness of this method is over-reliance on theextracted image features in the sampling window, face recongnition based onClass-Specific and HMM is researched in this dissertation. A concept of Class-Specificis introduced. Class-Specific Parameterization and Class-Specific HMM Algorithm areexplored. Obtaining results of Experiments on the ORL database, the FERET databaseand YALE database and the CAS-PEAL-R1illumination and expression database, andanalysing the effect of the algorithm and the comparative results, we can see that facerecongnition based on Class-Specific HMM is feasible.